Journal of Non-Crystalline Solids 355 (2009) 2571–2577

Contents lists available at ScienceDirect

Journal of Non-Crystalline Solids

journal homepage: www.elsevier .com/ locate/ jnoncrysol

Composition and structure dependence of the properties of lithiumborophosphate glasses showing boron anomaly

Francisco Muñoz a,*, Lionel Montagne b, Luis Pascual a, Alicia Durán a

a Instituto de Cerámica y Vidrio (CSIC), Kelsen 5, 28049 Madrid, Spainb Unité de Catalyse et Chimie du Solide, Université des Sciences et Technologies de Lille, Ecole Nationale Supérieure de Chimie de Lille, 59655 Villeneuve d’Ascq, France

a r t i c l e i n f o

Article history:Received 29 August 2008Received in revised form 9 September 2009Available online 21 October 2009

PACS:81.05.Kf82.45.Gj82.47.Aa82.56.-b

Keywords:Fast ion conductionGlassesMicrocrystallinityBoratesPhosphatesNMR, MAS-NMR and NQRShort-range order

0022-3093/$ - see front matter � 2009 Elsevier B.V. Adoi:10.1016/j.jnoncrysol.2009.09.013

* Corresponding author. Tel.: +34 917355840; fax:E-mail address: fmunoz@icv.csic.es (F. Muñoz).

a b s t r a c t

This work presents a study on the structure, microstructure and properties of 50Li2O�xB2O3�(50 � x)P2O5

glasses. The structure has been studied through NMR spectroscopy and the microstructure by TEM. Theproperties of the glasses are discussed according to their structure and microstructural features. Theintroduction of boron produces new linkages between phosphate chains through P–O–B bonds, whoseamount increases with boron incorporation; at the same time, a depolymerisation of the phosphatechains into Q1-type phosphate units takes place. The introduction of boron produces an increase in Tg

together with a decrease in the molar volume. The room temperature electrical conductivity increaseswith boron content as well. However, B2O3 contents higher than 20 mol% lead to crystallisation of lithiumorthophosphate which contributed to hinder ionic conduction of the glasses.

� 2009 Elsevier B.V. All rights reserved.

1. Introduction

The development of solid electrolytes for lithium secondarybatteries has become an interesting topic for many researchers ofthe glass community in the exploration of novel technologicalapplications of glassy materials [1–5]. The use of solid materialsas electrolyte components of the battery could avoid several draw-backs like those related to contamination of the electrode materialsand explosions when using liquid electrolytes [6,7]. The generalrequirements that solid electrolytes must fulfil are: a high ionicconductivity of lithium ions at room temperature, negligible elec-tronic contribution together with chemical and mechanical stabil-ity within the working potential range, and compatibility with theelectrode materials.

Nowadays, the most promising glassy electrolytes which havealready been tested in rechargeable battery devices belong to sys-tems of composition LiI–LiF–Li2S–Li2O–P2O5–P2S5 with many pos-sible variations [8,9]. Glasses of these systems possessconductivities as high as 10�3 S cm�1 at 25 �C, but they usually

ll rights reserved.

+34 917355843.

need of special synthesis conditions due to their high hygroscopic-ity. On the other hand, the high volatility of most of the compo-nents provokes difficulties to reproduce compositions.

Another approach which also yields promising results is the useof glass–ceramic electrolytes, i.e. materials with compositions ofthe systems Li2O–Al2O3–GeO2–P2O5 [10,11] or Li2O–Al2O3–TiO2–P2O5 [12]. Although they present a high chemical and thermal sta-bility, the processing of the anode/glass–ceramic/cathode devicescould be a source of additional problems such as complete elimina-tion of mechanical stresses.

From the point of view of the chemical interaction between theelectrode materials and electrolyte, the wettability of the glass toform stable interfaces is a crucial issue that glassy materials couldbest fulfil.

Lithium borophosphate glasses were first studied by Magistriset al. [13], who demonstrated that substitution of B2O3 for P2O5

leads to an increase in the room temperature ionic conductivityfor a maximum ratio [B2O3]/([ P2O5] + [ B2O3]) of around 0.5. Otherrecent studies on the conductivity of Li2O–B2O3–P2O5 glasses, withup to 0.47 molar fraction of Li2O, have been focused on their pos-sible application as thin-film solid electrolytes [14–16]. Cho et al.showed that the addition of P2O5 to lithium borate glasses de-

2572 F. Muñoz et al. / Journal of Non-Crystalline Solids 355 (2009) 2571–2577

creases the room temperature conductivity with a maximum valueof 1.6 � 10�7 S cm�1 [15]. The authors interpreted the decrease inconductivity by the formation of fourfold coordinated boron atomsand the consequent decrease in BO3 groups. This is, in fact, oppo-site to the assumption made by other authors that formation ofthe fourfold coordinated boron increases the conductivity becauseLi+ ions become weaker linked to BO4 than to BO3 [17].

The purpose of this work has been to relate the behaviour of theproperties of glasses with composition 50Li2O�xB2O3�(50 � x)P2O5,i.e. Tg, density, molar volume and ionic conductivity, withthe B2O3 content as well as with the structure and microstructureof the glasses, which have been followed through X-ray Diffrac-tion, Nuclear Magnetic Resonance and Transmission ElectronMicroscopy.

2. Experimental

2.1. Glass melting

Lithium borophosphate glasses with composition 50Li2O�x-B2O3�(50 � x)P2O5 (x = 2–25 mol%), have been obtained by conven-tional melt-quenching procedure. Reagent grade raw materialsLi2CO3 (Aldrich, 99%), (NH4)2HPO4 (Merck, 99%) and v-B2O3 (B2O3

previously melted at 800 �C) were mixed in stoichiometricamounts and the batches were calcined in porcelain crucibles upto 450 �C, in an electric furnace, and then melted during 1 h at tem-peratures ranging from 800 �C to 900 �C depending on composi-tion. The melts were poured onto brass moulds and annealedslightly above their glass transition temperature.

2.2. Characterisation of the glasses

Chemical analysis of the glasses was performed through Induc-tively Coupled Plasma-Emission Spectrometry (P2O5 and Al2O3) ina Thermo Jarrel Ash IRIS Advantage equipment, Flame Photometry(Li2O) in a Perkin–Elmer 2100 instrument and gravimetric and vol-umetric analysis to determine SiO2 and B2O3 contents, respectively.Silica and alumina were determined for some glasses in order tocheck possible contamination from the crucible. A total amountof less than 1 wt% for the sum of SiO2 and Al2O3 was found andthen not relevant for the glass properties. The uncertainties ofthe chemical analyses were ±0.3 wt%, 0.1 wt% and 0.5 wt% forLi2O, B2O3 and P2O5, respectively, and the total error is taken as±2 wt%.

Glass transition temperature has been determined by Differen-tial Thermal Analysis (DTA) in a SEIKO EXSTAR6000, TG-DTA6300,analyser using powdered samples in a platinum crucible, understatic air and a constant heating rate of 10 K min�1. Tg values areobtained at the onset of the endothermic effect of the DTA patternand the estimated error in the determination of Tg is assumed to bewithin ±2 K.

X-ray Diffraction (XRD) analysis of the glasses was carried outwith a D-5000 Siemens diffractometer using monochromatic CuKa radiation (1.5418 Å).

The density of the glasses was measured by helium pycnometryin a Quantachrome Corp. multipycnometer on bulk samples. Themolar volume of the glasses (Vm) has been calculated from densitymeasurements by using the relation Vm = M/d (in cm3 mol�1),being M the molecular mass, and d the density of the glasses.

Transmission Electron Microscopy (TEM) was performed in aHitachi H7100 using the carbon replica method. The samples werechemically etched with a 5 vol.% HF solution during 20 s.

Electrical conductivity measurements were performed by Elec-trochemical Impedance Spectroscopy (EIS) in a Gamry REF600impedance analyser, within the frequency range from 10 Hz to

1 MHz at temperatures between 25 and 120 �C, with an appliedvoltage of 0.5 V. The samples were cut into discs of 1–2 mm inthickness and �10 mm in diameter and gold electrodes were sput-tered on both faces as contacts for electrical measurements. Theelectrical conductivity (r) is determined, for each temperature,through the resistance value (R) read at the low frequency intersec-tion of the semicircle with the x-axis in the Nyquist plots and usingthe sample geometric factor (e/A; e = thickness, A = electrode area)through the equation r = e/(R � A). The error in the determination ofthe conductivity is estimated to be less than 10%. The experimentaldata are then represented as a function of the reciprocal tempera-ture and have been fitted to an Arrhenius equation of the type:

r ¼ r0 exp�Ea=RT; ð1Þ

where r0 and R are the pre-exponential factor and gases constant,respectively, and Ea is the activation energy for conduction.

31P MAS (Magic Angle Spinning) NMR (Nuclear Magnetic Reso-nance) spectra were recorded on a Bruker ASX 400 spectrometeroperating at 161.96 MHz (9.4 T). The pulse length was 2.5 ls and60 s delay time was used (sufficient to enable relaxation). A totalnumber of 128 scans were accumulated with a spinning rate of10 kHz. The 31P spectra were fitted to Gaussian functions, in accor-dance with the chemical shift distribution of the amorphous state[18]. The precision of the relative component determination was±5%. Solid (NH4)H2PO4 was used as secondary reference with achemical shift 0.82 ppm with respect to H3PO4 (85%).

11B MAS NMR spectra were recorded on a Bruker ASX 400 spec-trometer operating at 128.38 MHz (9.4 T). The pulse length was1.5 ls and 3 s delay time was used. A total number of 512 scanswere accumulated under a spinning rate of 10 kHz. A solution ofBF3�Et2O was used as reference.

7Li MAS NMR was performed on a Bruker ASX 400 spectrometeroperating at 155.51 MHz (9.4 T). The pulse length was 2 ls and 2 sdelay time was used. A total number of 256 scans were accumu-lated under a spinning rate of 10 kHz. Solid LiCl was used as sec-ondary reference with chemical shift of �1.06 with respect toLiCl 1 M solution.

3. Results

Table 1 shows the nominal and analysed compositions of theglasses as well as Al2O3 and SiO2 contents for 50Li and 50Li10Bsamples, in mol%. All prepared glasses were homogeneous andtransparent except 50Li25. This appeared with milky white aspectleading to opaque glass, suggesting that a liquid–liquid phase sep-aration and/or crystallisation occurred. Glasses with higher than25 mol% B2O3, i.e. 35, clearly crystallised after quenching. Fig. 1presents the XRD patterns of the 20, 25 and 35 mol% B2O3 contain-ing samples. The XRD pattern of 50Li20B glass presents broad crys-tallisation peaks, clearly defined for further boron additions in50Li25B and 50Li35B glasses, and identified as crystallisation ofLi3PO4 (PDF file No. 15-0760).

Table 2 gathers the glass transition temperatures (Tg), the den-sity and molar volume (Vm) and the activation energy for conduc-tion (Ea) of the borophosphate glasses.

Fig. 2 depicts the variation of Tg of the borophosphate glasses asa function of the boron oxide molar content. The glass transitiontemperature increases within the whole range of B2O3 contents,as already observed by Magistris et al. [13]. It shows an approxi-mate linear behaviour between LiPO3 (0 mol% B2O3) and 50Li20Bglasses. This increase in Tg is attributed to the increased cross-link-ing of the glass network between phosphate chains through P–O–Bbonds, as will be shown hereafter.

Fig. 3 presents the density and the molar volume of the glassesas a function of the B2O3 content. Glass density increases from 0 to

Table 1Nominal an analysed compositions of the borophosphate glasses, in mol%.

Glass Li2O nominal Li2O analysed B2O3 nominal B2O3 analysed P2O5 nominal P2O5 analysed Al2O3 SiO2

50Li 50 49.1 0 0 50 50.3 0.05 0.5550Li2B 50 49.2 2 2 48 48.8 – –50Li5B 50 49.4 5 4.7 45 46.0 – –50Li10B 50 49.7 10 9.6 40 40.1 0.03 0.5850Li15B 50 49.9 15 14.7 35 35.4 – –50Li20B 50 48.2 20 20.1 30 31.2 – –50Li25B 50 48.5 25 25.5 25 26.0 – –

10 20 30 40 50 60 70angle 2 (º)θ

50Li20B

50Li25B

50Li35B

Fig. 1. X-ray Diffraction patterns of the glasses with 20, 25 and 35 mol% B2O3,presenting crystallisation of Li3PO4.

Table 2Glass transition temperature (Tg), density, molar volume and activation energy forconduction (Ea) of lithium borophosphate glasses.

Glass Tg

(±2 �C)Density(±0.01 g cm�3)

Vm

(cm3 mol�1)Ea

(±0.01 eV)

50Li 311 2.34 36.6 0.8350Li2B 327 2.33 36.2 0.8450Li5B 339 2.38 34.5 0.7550Li10B 360 2.40 32.8 0.6750Li15B 382 2.45 30.7 0.6750Li20B 407 2.44 29.2 0.6350Li25B n. d.a 2.41 28.2 0.67

a Not determined.

0 5 10 15 20mol % B2O3

300

320

340

360

380

400

420

T g (º

C)

Fig. 2. Variation of the glass transition temperature of the borophosphate glasses asa function of the B2O3 content. Line is drawn as a guide for the eyes.

0 5 10 15 20 25mol % B2O3

2.25

2.30

2.35

2.40

2.45

2.50

dens

ity (g

cm

-3)

25

30

35

40

Vm (cm

3 mol -1)

density

Vm

Fig. 3. Density and molar volume (Vm) of the borophosphate glasses as a function ofthe B2O3 content. Lines are drawn as a guide for the eyes.

F. Muñoz et al. / Journal of Non-Crystalline Solids 355 (2009) 2571–2577 2573

15 mol% B2O3, then it slightly decreases for higher boron contents.On the opposite, molar volume shows a continuous decrease be-tween LiPO3 and 50Li25B glass compositions. The decrease in mo-lar volume indicates a higher reticulation of the glass networkwithin the whole range of compositions, this being responsiblefor the increase of the glass transition temperature. Fig. 4 depictsthe logarithm of the electrical conductivity as a function of the re-ciprocal absolute temperature for the 50Li2O�xB2O3�(50 � x)P2O5

glasses. Activation energy for conduction (see Table 2) decreasesfor B2O3 contents up to 20 mol%, remaining approximately

constant within the error limits for further boron additions. Fig. 5plots the Log of conductivity at 25 �C, extrapolated from the Arrhe-nius fits, as a function of the B2O3 content in the glasses. The roomtemperature conductivity, Log r25 �C, increases with B2O3 up to20 mol%, then slightly decreases for 50Li25B glass. Similar resultshave been observed by Magistris et al. [13], Cho et al. [15] and re-cently by Money et al. [19].

Fig. 6 presents the 31P MAS NMR spectra of the 50Li2O�x-B2O3�(50 � x)P2O5 glasses from 2 to 25 mol% B2O3. For the 50Li2B

2.4 2.6 2.8 3.0 3.2 3.4

103/T (K-1)

-9

-8

-7

-6

-5

-4

-3lo

g

( in

S c

m-1

)σ

σ

50Li2BLiPO3

50Li5B50Li10B

50Li25B

50Li15B50Li20B

Fig. 4. Log of conductivity for the 50Li2O�xB2O3�(50 � x)P2O5 glasses as a function ofthe reciprocal temperature. Experimental data points have been fitted to anArrhenius type equation through the least squares method obtaining correlationcoefficients of 0.998 (50LiPO3), 0.999 (50Li2B), 0.999 (50Li5B), 0.999 (50Li10B),0.999 (50Li15B), 0.999 (50Li20B) and 0.999 (50Li25B), from which Log r at 298 Kand Ea have been calculated.

0 5 10 15 20 25mol % B2O3

-9.2

-8.8

-8.4

-8.0

-7.6

-7.2

-6.8

Log

( in

S c

m-1

), 25

°Cσ

Fig. 5. Logarithm of conductivity at 298 K as a function of the molar percentage ofB2O3 in the lithium borophosphate glasses. The values have been calculated byextrapolation using the Arrhenius fits in Fig. 4 at 298 K. The error bars are withinthe size of the symbol and line has been drawn as a guide for the eyes.

-60-50-40-30-20-1001020

chemical shift 31P (ppm)

Inte

nsity

(a. u

.)

50Li2B

50Li5B

50Li10B

50Li20B

50Li25B

Q2

Q1 P-O-BQ0

Fig. 6. 31P MAS NMR spectra of the 50Li2O�xB2O3�(50 � x)P2O5 glasses with x = 2–25 mol% B2O3.

-20-1001020chemical shift 11B (ppm)

Inte

nsity

(a. u

.)

50Li2B

50Li5B

50Li10B

50Li20B

50Li25B

Fig. 7. 11B MAS NMR spectra of 50Li2O�xB2O3�(50 � x)P2O5 glasses from x = 2–25 mol% B2O3.

2574 F. Muñoz et al. / Journal of Non-Crystalline Solids 355 (2009) 2571–2577

glass, the decomposition of the spectrum needed three Gaussianbands at �4.4, �14.6 and �22.7 ppm. Phosphate glasses are builtup of PO4 tetrahedra, which are denominated Qn groups, as fromLippmaa et al. definition [20], where n denotes the number ofbridging oxygen atoms which are shared by two neighbouringphosphorous. In a metaphosphate glass, all phosphate groups areQ2, having two bridging oxygens and two non-bridging equivalentones, which gives rise to a network formed by long polymericchains based on PO4 tetrahedra. The decrease in the P2O5 contentresults in the formation of Q1 (pyrophosphate) and Q0 (orthophos-phate) groups. The resonances at �4.4 and �22.7 are assigned toQ1 and Q2 sites [21], respectively, while the resonance at�14.55 ppm is identified as phosphorous atoms belonging to P–O–B bonds, which result from the substitution of B2O3 for P2O5 inthe LiPO3 glass, as already observed in borophosphate [22] and inP2O5-containing borosilicate glasses as well [23]. The increase in

B2O3 content in the lithium borophosphate glasses leads to the in-crease in the intensity of the Q1 and P–O–B groups and the de-crease in the Q2 phosphate units. A shift downfield of the threeresonances is also observed with addition of B2O3 as already ob-served in phosphate glasses due to shortening of the phosphatechains. In addition, the 31P NMR spectra of the 20 and 25 mol%B2O3 glasses show a small but sharp peak at +9.5 ppm, which cor-responds to crystallisation of Li3PO4 [24].

Fig. 7 depicts the 11B MAS NMR spectra of the 50Li2O�x-B2O3�(50 � x)P2O5 glasses from 5 to 25 mol% B2O3. They show res-onances typical for the presence of BO4 as well as BO3 groups for

-20-1001020chemical shift 7Li (ppm)

Inte

nsity

(a. u

.)

50Li2B

50Li20B

Fig. 8. 7Li MAS NMR spectra of the glasses containing 2 and 20 mol% B2O3.

F. Muñoz et al. / Journal of Non-Crystalline Solids 355 (2009) 2571–2577 2575

high boron-containing glasses. The BO4 resonances exhibit Gauss-ian/Lorentzian lineshape, located at �3.4, �2.5 and �1.7 ppm thatare attributed to B(OP)4, B(OP)3(OB) and B(OP)2(OB)2 species,respectively [22]. The BO4 resonance at �3.4 ppm is observed forthe glasses up to 10 mol% B2O3 with a small shift downfield withincreasing boron content, but all the boron atoms remain fourfoldcoordinated in the glasses up to around 20 mol% B2O3. The broadresonance with a quadrupolar lineshape centred at ca. 15 ppmfor B2O3 content P20 mol%, is assigned to trigonal BO3 groups[22]. The increase in the boron content is reflected in the increasein the amount of BO3 trigonal groups respecting to BO4 ones.

Tian et al. observed that for a constant 45 mol% Li2O-containingglasses, N4 (defined as the ratio between BO4 groups and totalamount of borate groups) first increases up to 15 mol% B2O3 thendecreasing for further boron addition [17]. The results found inthe present work show instead that a change in the coordinationenvironment of boron atoms takes place at around 20 mol% B2O3.For a constant 50 mol% Li2O composition, BO4 units are the pre-dominant species from 0 up to �20 mol% B2O3, and then N4 pro-gressively decreases with further boron additions.

The 7Li MAS NMR spectra of the 50Li2O�2B2O3�48P2O5 (50Li2B)and 50Li2O�20B2O3�30P2O5 (50Li20B) glasses are shown in Fig. 8.They both present a single resonance centred at �0.75 ppm, butnarrower for 50Li20B. It is characteristic to four-coordinated Li+,as observed in LiPO3 glass [25].

Fig. 9. TEM microphotographs of the g

Fig. 9 presents the TEM microphotographs of the glasses with20 (a) and 25 (b) mol% B2O3, 50Li20B and 50Li25B, respectively.Glasses with compositions up to 15 mol% B2O3 were all homoge-nous. Small heterogeneities can be observed for borophosphateglass with 20 mol% B2O3 (Fig. 9(a)). 50Li25B glass (Fig. 9(b)) pre-sented non-spherical defects resulting from the crystallisation ofLi3PO4, as proved by XRD.

4. Discussion

The NMR results show two important structural features as theB2O3 content increases in the lithium borophosphate glasses:

i. P–O–B bonds are present at very low B2O3 content and theirrelative amount increases within the whole range of B2O3

contents.ii. BO4 tetrahedral borate groups are predominant, �100%, up

to 15 mol% B2O3. Then, proportion of BO3 units increases.

In addition, 31P NMR spectra (Fig. 6) showed that the reticula-tion effect is accompanied by the formation of Q1 terminal phos-phate groups, or pyrophosphate ones, a sign of the phosphatenetwork depolymerisation. However, Q1 phosphate units mightalso be bonded to borate ones through P–O–B linkages, so that theycannot be considered as completely separated structural entities.The widening of the resonance bands does not allow to clearly dis-tinguish how many, and in which proportion, the different types ofphosphate species (Q2 and Q1 units) are bonded to either BO3 orBO4 boron groups. Therefore, a transition from a phosphate to aborophosphate glass network is expected to occur. Finally, the Q1

phosphate units break down into Q0 sites that crystallise as Li3PO4,as shown by XRD for compositions with 20 mol% 6B2O3.

As boron substitutes for phosphorous, the amount of P–O–Bbonds increases and, from 20 mol% B2O3 glass, the BO3 groups ap-pear, giving rise to new linkages of the type B–(OP)3 together withB–O–B ones.

From these structural considerations, the observed increase inTg (Fig. 1) is attributed to the formation of both tetrahedral (BO4)

lasses with 20 and 25 mol% B2O3.

2576 F. Muñoz et al. / Journal of Non-Crystalline Solids 355 (2009) 2571–2577

or trigonal (BO3) groups linked to phosphate chains, that increasethe network polymerisation through P–O–B bonds. The main rea-son for the increase in Tg comes from the reticulation producedby the P–O–B bonds which link together the phosphate chains. Thiseffect is expected to be more important for BO4 which form B–(OP)4 linkages, than for BO3 groups. Furthermore, the change inboron coordination, from BO4 to BO3, is also reflected in the de-crease in density for boron contents higher than 15 mol%, thoughthe molar volume of the glasses continuously decreases withinthe whole range of compositions. For glasses with 20 and25 mol% B2O3, BO3 trigonal units are clearly present in the struc-ture of the glasses but, at the same time, phase separation occursand crystallization of Li3PO4. Phase separation as well as crystalli-sation might also play an important role contributing to alter thedensity of the samples for B2O3 contents higher than 20 mol%when those phenomena become more important, while the molarvolume of the continues decreasing.

This structural model is also used to explain the behaviour ofthe room temperature conductivity of the glasses when B2O3 isintroduced in lithium metaphosphate glasses. From 2 to 10 mol%B2O3, the Log of conductivity increases in an approximately linearmanner, then showing a slower increasing rate between 10 and20 mol% B2O3 (Fig. 5). The 50Li25B glass presents a lower Log r25 �C

than the one of 50Li20B. This conductivity behaviour may be sep-arated within three different regions depending on the structuraland microstructural features of the glass network observed:

Region 1: The increase in conductivity from 2 to 10 mol% B2O3 isdue to the formation of B–(OP)4 tetrahedra within a borophosphateglass network. Part of Li+ ions behave as charge compensators ofthe BO4 groups, thus increasing their mobility with respect to thosein a pure phosphate network where all Li+ behave as modifier ions.

Region 2: The lower increasing rate of conductivity from 10 to20 mol% B2O3 is due to the decrease in the B–(OP)4 bonds andthe increase in the newly formed B–(OP,B)3. Formation of trigonalBO3 will contribute to decrease the amount of Li+ as charge com-pensator, thus counteracting the increase in conductivity due tothe BO4 units. The higher the amount of BO3 the lower the increasein conductivity.

Region 3: Decrease in conductivity for 25 mol% B2O3, due toLi3PO4 crystallisation. Lithium orthophosphate will block Li+ ionsas the crystalline phase is developed within the glass matrix. Crys-tallisation of Li3PO4 might also hinder mobility of Li+ ions in50Li20B and, for these higher B2O3 compositions, the phase separa-tion phenomenon producing a heterogeneous microstructure (seeFig. 9) is also responsible of a more difficult conduction throughthe sample.

Money et al. have recently reported the Log of conductivity at110 �C of 50Li2O�xB2O3�(50 � x)P2O5 glasses as a function of theB2O3 content [19]. They attributed the increase in conductivity be-tween 10 and 20 mol% B2O3 to BPO4 units which form new conduc-tion pathway for lithium ions reducing the activation energy forion migration. If BPO4 units were formed in borophosphate glasses,31P NMR should show a resonance band at �30 ppm [26], which isactually not seen. Furthermore, in their study, Money et al. [19]proposed that the highest conductivity value observed for the25 mol% B2O3 composition could be attributed to ‘a dispersion ofLi3PO4 crystallites within the glass matrix’. The present workshows that Li3PO4 starts crystallising from 20 mol% B2O3, and theglass fully crystallises only for the 35 mol% B2O3 composition. Fora 25 mol% B2O3 sample, there is strong phase separation (seeFig. 9), however, it is not easy to think on a continuous crystallinephase of Li3PO4. As discussed above, we can conclude that the in-crease in the conductivity of the lithium borophosphate glasses isspecifically due to the new structural units formed in the glassymatrix through P–O–B bonds, which modify the bonding of Li+ ionsas charge compensators in tetrahedral BO4 units, and not to Li3PO4

crystallites or BPO4 species. The non-linear changes in propertiesobserved for the lithium borophosphate glasses, such as densityand Log r25 �C, resemble those provoked in alkali borate glassesdue to the so-called boron anomaly [27]. Increasing alkali contentleads to anomalous changes in the properties of the borate glasses,related to the transformation of BO3 groupings into BO4 ones. Thenon-linear behaviour observed in the properties takes place for themaximum amount of BO4 tetrahedra, which increases the reticula-tion of the glass network. In the present case, a continuous increas-ing of the B2O3 content produces a modification of the glassproperties while all the boron species are BO4 units. Once BO4 startto transform into BO3 groups, the critical point in properties is at-tained (deviation density and Log r maxima), then showing achange in their linear behaviour. It is then worth mentioning itas a particular case of the boron anomaly.

5. Conclusions

A structure–properties relationship has been achieved forglasses of composition 50Li2O�xB2O3�(1 � x)P2O5. The glasses wereobtained transparent for B2O3 contents between 0 and 20 mol%.Glasses with B2O3 > 20 mol% presented crystallisation of Li3PO4,as well as phase separation, becoming fully crystallised in the caseof 35 mol% B2O3.

The borophosphate glass network is built up of Q1 and Q2-typephosphate groups linked to BO4 and BO3 ones through P–O–Bbonds. The higher the boron content the higher the P–O–B bondsand Q1 phosphate groups. BO4 tetrahedra are predominant up to20 mol% B2O3. Addition of boron into lithium phosphate glassesgives rise to an increase in Tg and decrease in the molar volumeof the glasses, due to the increased reticulation of the networkthrough the new P–O–B linkages formed. The room temperatureconductivity of the glasses increases with boron addition up to20 mol% B2O3 due to (i) formation of BO4 units, predominantly overBO3 ones, which allow Li+ ions behave as charge compensators;and (ii) decrease of the Li–Li average distance through the decreasein molar volume. For B2O3 >20 mol%, ionic conduction is hinderedas a consequence of the progressively increased amount of BO3 bo-rate units and, particularly, crystallisation of Li3PO4 and phase sep-aration phenomena. The non-linear behaviour of the glassproperties of lithium borophosphate glasses is interpreted in termsof a boron anomaly occurring when BO4 tetrahedra transform intoBO3 groups after B2O3 substitutes for P2O5.

Acknowledgements

The authors thank financial support from the ENERVID project(CICYT-MAT2006-4375) of the Spanish National Materials Pro-gramme. F. Muñoz acknowledges the I3P contract from CSIC. Theauthors are also grateful to E. Peiteado for her assistance in thepreparation and characterisation of the glasses.

References

[1] A.G. Ritchie, J. Power Sources 136 (2004) 285.[2] R. Komiya, A. Hayashi, H. Morimoto, M. Tatsumisago, T. Minami, Solid State

Ionics 140 (2001) 83.[3] R.C. Agrawal, R.K. Gupta, J. Mat, Science 34 (1999) 1131.[4] H. Morimoto, H. Yamashita, M. Tatsumisago, T. Minami, J. Am. Ceram. Soc. 82

(5) (1999) 1352.[5] M. Duclot, J.-L. Souquet, J. Power Sources 97&98 (2001) 610.[6] B.B. Owens, J. Power Sources 90 (2000) 2.[7] J.-M. Tarascon, M. Armand, Nature 414 (2001) 359.[8] T. Minami, A. Hayashi, M. Tatsumisago, Solid State Ionics 136&137 (2000)

1015.[9] Y. Kowada, M. Tatsumisago, T. Minami, H. Adachi, J. Non-Cryst. Solids 354

(2008) 360.[10] J. Fu, Solid State Ionics 104 (1997) 191.[11] X. Xu, Z. Wen, X. Wu, X. Yang, Z. Gu, J. Am. Ceram. Soc. 90 (9) (2007) 2802.

F. Muñoz et al. / Journal of Non-Crystalline Solids 355 (2009) 2571–2577 2577

[12] I. Abrahams, E. Hadzifejzovic, Solid State Ionics 134 (2000) 249.[13] A. Magistris, G. Chiodelli, J. Power Sources 14 (1985) 87.[14] S. Lee, J. Kim, D. Shin, Solid State Ionics 178 (2007) 375.[15] K.I. Cho, S.H. Lee, K.H. Cho, D.W. Shin, Y.K. Sun, J. Power Sources 163 (2006)

223.[16] K.-H. Cho, H.-J. You, Y.-S. Youn, J.-S. Kim, D.-W. Shin, Electrochim. Acta 52

(2006) 1571.[17] F. Tian, L. Pan, J. Non-Cryst. Solids 112 (1989) 142.[18] D. Massiot, F. Fayon, M. Capron, I. King, S. Le Calvé, B. Alonso, J.O. Durand, B.

Bujoli, Z. Gan, G. Hoatson, Magn. Reson. Chem. 40 (2002) 70.[19] B.K. Money, K. Hariharan, Solid State Ionics 179 (2008) 1273.[20] E. Lippmaa, M. Maegi, A. Samoson, G. Engelhardt, A. Grimmer, J. Am. Chem.

Soc. 102 (1980) 4889.

[21] L. Wüllen, H. Eckert, G. Schwering, Chem. Mater. 12 (2000) 1840.[22] M. Zeyer-Düsterer, L. Montagne, G. Palavit, C. Jäger, Solid State NMR 27 (2005)

50.[23] F. Muñoz, L. Montagne, L. Delevoye, A. Durán, L. Pascual, S. Crystol, J.-F. Paul, J.

Non-Cryst. Solids 352 (2006) 2958.[24] R.J. Kirkpatrick, R.K. Brow, Solid State Nucl. Magn. Reson. 5 (1995) 9.[25] T.M. Alam, S. Conzone, R.K. Brow, T.J. Boyle, J. Non-Cryst. Solids 258 (1999)

140.[26] M. Zeyer-Düsterer, L. Montagne, G. Palavit, C. Jäger, Solid State Nucl. Magn.

Reson. 27 (2005) 50.[27] W. Vogel, in: Glass Chemistry, 2nd Ed., Springer-Verlag, 1994, p. 138.